Bearing Fault Detection Based on Improved GWO to Optimize Multi-Stable Stochastic Resonance Parameters

For the problem that the traditional stochastic resonance system cannot adaptively adjust the system parameters in the bearing fault signal detection. In this paper, a bearing fault signal detection method based on improved gray Wolf algorithm is proposed for adaptive optimization of multi-stable stochastic resonance parameters Firstly, the best point set theory was used to initialize the grey wolf population to improve the diversity of the population. An exponential convergence factor is introduced to balance the global search and local development of the algorithm, and an adaptive position update strategy is introduced to improve the accuracy of the algorithm. Then, based on the multi-stable stochastic resonance system model, the structure parameters of the multi-stable stochastic resonance were optimized by adaptive improvement of the grey wolf algorithm, so as to enhance the fault signals and realize the detection of bearing fault signals. The experimental results shown that the proposed method has fast convergence speed, high output signal-to-noise ratio of the multi-stable stochastic resonance system, and accurate and efficient fault detection performance. The method lays a foundation for the application of adaptive multi-stable stochastic resonance in engineering practice.